Connect public, paid and private patent data with Google Patents Public Datasets

Instruction cache system for implementing programs having non-sequential instructions and method of implementing same

Download PDF

Info

Publication number
US5701435A
US5701435A US08063845 US6384593A US5701435A US 5701435 A US5701435 A US 5701435A US 08063845 US08063845 US 08063845 US 6384593 A US6384593 A US 6384593A US 5701435 A US5701435 A US 5701435A
Authority
US
Grant status
Grant
Patent type
Prior art keywords
instruction
cache
instructions
branch
program
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Lifetime
Application number
US08063845
Inventor
Chi-Hung Chi
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Intellectual Ventures II LLC
Original Assignee
Philips Electronics North America Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Grant date

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06FELECTRICAL DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for programme control, e.g. control unit
    • G06F9/06Arrangements for programme control, e.g. control unit using stored programme, i.e. using internal store of processing equipment to receive and retain programme
    • G06F9/30Arrangements for executing machine-instructions, e.g. instruction decode
    • G06F9/38Concurrent instruction execution, e.g. pipeline, look ahead
    • G06F9/3802Instruction prefetching
    • G06F9/3808Instruction prefetching for instruction reuse, e.g. trace cache, branch target cache
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06FELECTRICAL DIGITAL DATA PROCESSING
    • G06F12/00Accessing, addressing or allocating within memory systems or architectures
    • G06F12/02Addressing or allocation; Relocation
    • G06F12/08Addressing or allocation; Relocation in hierarchically structured memory systems, e.g. virtual memory systems
    • G06F12/0802Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches
    • G06F12/0862Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches with prefetch
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06FELECTRICAL DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for programme control, e.g. control unit
    • G06F9/06Arrangements for programme control, e.g. control unit using stored programme, i.e. using internal store of processing equipment to receive and retain programme
    • G06F9/30Arrangements for executing machine-instructions, e.g. instruction decode
    • G06F9/38Concurrent instruction execution, e.g. pipeline, look ahead
    • G06F9/3802Instruction prefetching
    • G06F9/3808Instruction prefetching for instruction reuse, e.g. trace cache, branch target cache
    • G06F9/381Loop buffering
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06FELECTRICAL DIGITAL DATA PROCESSING
    • G06F12/00Accessing, addressing or allocating within memory systems or architectures
    • G06F12/02Addressing or allocation; Relocation
    • G06F12/08Addressing or allocation; Relocation in hierarchically structured memory systems, e.g. virtual memory systems
    • G06F12/0802Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches
    • G06F12/0875Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches with dedicated cache, e.g. instruction or stack
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06FELECTRICAL DIGITAL DATA PROCESSING
    • G06F2212/00Indexing scheme relating to accessing, addressing or allocation within memory systems or architectures
    • G06F2212/60Details of cache memory
    • G06F2212/6028Prefetching based on hints or prefetch instructions

Abstract

A system for increasing the speed and efficiency of instruction execution by a computer processing system. An instruction cache is provided to receive a minor number of stored instructions for execution by the computer processing system. The instructions are prefetched and returned in cache based upon an analysis of instructions which are in the cache pending execution. Target instructions of branch instructions may be prefetched as a result of the analysis of a branch instruction pending in the cache. Other instructions may be retained in cache when they are tagged as being likely to be reused.

Description

This is a continuation of application Ser. No. 07/500,612, filed Mar. 27, 1990 now abandoned.

This invention relates to computer processing systems which make use of instruction cache memories to increase instruction access speed. Specifically, a method and apparatus for reducing instruction cache misses without increasing instruction cache pollution is described.

Stored memory programs are implemented in computer processing systems which are executed by sequential instruction line execution. The retrieval of a line of instructions from memory necessarily requires access to the system bus slowing the overall computation time for the computer.

To alleviate the bus overhead in line by line program execution, instruction caches have been used which permit prefetching of groups of lines of instructions during some cycle(s) of the memory bus. The processor may then reference a line of instruction from the cache without regard for the bus availability.

Some programs are not entirely sequential in that they have branch instructions or loop instructions which transfer execution to an earlier instruction. If instructions are prefetched on a sequential basis, a cache miss might occur when the next instruction is a target instruction of branch instructions. The cache must then use some bus cycle(s) to acquire the referenced instruction.

The prefetching of instructions for execution may also present the additional problem of cache pollution, wherein lines of instructions are prefetched into the cache and never used. The result is wasted cache memory space and the increased potential for a cache miss due to non-referenced instructions being stored in the cache.

One solution to this dilemma is a cache prefetching operation based on a one line look ahead. When a cache miss occurs, the reference line and next consecutive line of instructions are prefetched in the next available bus access. This scheme does not verify whether the next line will actually be used, thus presenting the possibility of polluting the cache rather than enhancing its performance.

SUMMARY OF THE INVENTION

It is an object of this invention to provide efficient prefetching of instructions from a program memory to an instruction cache.

It is yet another object of this invention to compile a computer program to avoid cache misses during execution of the program.

It is still another object of this invention to manage an instruction execution cache to avoid pollution of the cache with unused instructions.

These and other objects of the invention are provided by an instruction execution cache which prefetches during an available bus cycle a plurality of instructions from a main program storage memory. The fetched instructions are examined by a prefetch processor to determine the likelihood of a cache miss by the computer during sequential execution of the prefetched instructions.

This prediction of a cache miss is based upon determining whether or not an instruction such as a branch instruction has been loaded in the cache. Since these instruction types result in the program execution being non-sequential, forcing execution to a target instruction, the next instructions to be prefetched include the target instruction.

Additionally, when instructions are known to belong to a group of repetitively executed instructions, the portion of the instruction cache containing repetitively executed instructions is maintained frozen, i.e., in a state wherein they can be read and cannot be overwritten. Thus, it is possible to keep the beginning instructions of a program block which have a high probability of reuse in cache for repetitive use.

In carrying out the invention, explicit instructions may be inserted in the program to require a prefetch of the target instructions of a branch instruction in the next available bus cycle. An instruction to freeze and unfreeze instructions in cache may be inserted in the program to identify when these measures are to be taken.

DESCRIPTION OF THE FIGURES

FIG. 1 illustrates a cache memory interface between a program memory 13 and processing unit 11.

FIG. 2 illustrates the program organization in memory 13 which provides for explicit prefetching of target instructions before execution of a branch instruction.

FIG. 3 illustrates the program organization in memory 13 which provides for cache instruction freezing.

FIG. 4 illustrates the cache instruction contents during program execution having instructions which are frozen.

FIG. 5 illustrates a computing system architecture for implementing the cache instruction prefetching according to the instruction flow of FIGS. 2 and 3.

DESCRIPTION OF THE PREFERRED EMBODIMENT

Referring now to FIG. 1, there is shown an instruction processing unit (IPU) which is common to many computer processing system architectures. Although shown as a single operating system, it is clear and well known to use the instruction cache interface between a processing unit 11 and program memory 13 for parallel operating systems. Such instruction cache arrangements are demonstrated in the Convex Computer Corporation CI processor series literature.

The instruction cache 12 permits the processing unit 11 to prefetch a number of instruction for execution during an available bus cycle of the computer bus 10. The prefetched instructions are stored in the instruction cache 12, thus freeing the bus to be used for other tasks to be performed by the processing unit 11. By freeing up the bus 10 from instruction fetching operations, the overall processing speed may be improved by permitting the bus 10 to service other demands for the processing unit 11.

In a typical system operation, the processing unit 11 may fetch one or more lines of instructions from the memory 13. The lines of instruction in turn contain a plurality of words, each of which may correspond to a single sub-instruction for the line of instructions.

The fetched instructions located in the cache 12 are then sequentially processed by the processing unit 11. As most programs are written in terms of a series of blocks of instructions, having a single entry and exit point, a sequential prefetching of the instructions will often provide the necessary series of instruction in the cache 12 for execution.

During execution of the instructions by the processing unit 11, when the last word of a particular instruction line is referenced by the processing unit 11, preparatory to executing the instruction, the processing unit 11 checks for the next following line to determine whether or not it is in the instruction cache 12. A cache prefetch for the next line of instructions following the line containing the last word to be executed is issued if it is not in the instruction cache 12.

Certain conditional branch instructions alter the execution sequence of the processing unit 11. Thus, in the case of an "if " statement or other type of branch instruction, the next needed instruction is likely one which may have been previously fetched and which has since been discarded in the instruction cache 12.

In one embodiment of the present invention, the instruction cache 12 will not prefetch the next instruction from memory 13 when a branch instruction has been prefetched. However, if the instruction is an uncondition branch instruction, a prefetch bit may be set in the instruction cache 12, allowing the next consecutive instruction to be prefetched during the next available bus cycle.

Once a conditional branch instruction is executed, a bus cycle for bus 10 may be needed to fetch the particular target instruction. Thus, no prefetching is resumed in the instruction cache 12 until the branch instruction has been executed.

This particular scheme of inhibiting prefetching, when a conditional branch instruction is set to be executed, avoids the inevitable cache miss which will occur should a sequential instruction be fetched from the program memory 13 before a conditional branch instruction is executed, transferring control to its related target instruction which is probably not the next sequential instruction.

The foregoing prefetching conditions may be implemented by analylzing the program behavior in light of branch instructions, and other events which require the execution of instructions which are non-sequential. By anticipating that future references by the processing unit 11 to instructions of the stored program of memory 13 may not be sequential, it is possible to avoid polluting the instruction cache 12 with non-used instructions without incurring any additional risk of a cache miss 12 by interrupting the sequential prefetching of program instructions.

In implementing the foregoing technique, a conditional branch instruction may be encountered which changes the program execution sequence to another line perhaps ahead of those being executed. To avoid implicitly prefetching instructions, which immediately follow such a conditional branch instruction, the branch instruction is inspected and if it is a conditional branch, the cache prefetch bit is set to inhibit prefetching of subsequent branch instructions until the branch instruction is executed. Once the branch instruction is executed, and the processing unit 11 correctly identifies and references the target instruction from memory 13, prefetching may once again commence in the sequential manner until further branch instructions are encountered.

The foregoing techniques for deciding when to prefetch, on a sequential basis, instructions contained in a program memory 13, works well within blocks of instructions comprising the program. By recognizing when programs are not to be executed sequentially, the instruction cache may avoid a miss and avoid unnecessarily prefetching instructions which will not be used. This conserves bus cycle time, as well as polluting the instruction cache 12 with unused instructions.

Control over prefetching may also be effected by including explicit prefetch instructions at various points in the program, identifying to the cache and computer processor an instruction address to be executed and which is non-sequential.

FIG. 2 illustrates the memory 13 organization of a series of program instructions which are a part of a block beginning with instruction address 001. At address 010, there is a branch instruction, identifying a conditional target which is the beginning of the instruction block identified by address 001. Thus, sequential execution occurs for lines of instruction located at addresses 001 through 009.

Also shown in the block which begins with address 001 is an explicit prefetch instruction at address 009. The explicit prefetch instruction, when executed by the computer processor, will identify to the cache the instruction located at address 001, before the branch instruction is executed by the processor 11. Thus, by prefetching address 001, the target instruction for branch instruction at address 010 will be available in cache and avoid a miss before the execution of the branch instruction. Sequential prefetching may continue from address 001 at the next available bus cycle, permitting continued sequential prefetching of the instructions in memory 13.

Use of the explicit prefetch instruction is advantageous at the end of loops, wherein the program counter for the processor 11 would naturally jump back to the start of the loop. Branch instructions at the end of a loop are conditional, and, although a program instruction execution will eventually leave the loop, the higher probability is that the program instruction execution sequence will remain within the loop. Obviously, once the branch instruction 010 has been satisfied, and instruction 011 is to be executed, a miss will occur in the cache when the processor references the instruction 011. However, the miss penalty under this condition is far outweighed by prefetching the higher probability beginning loop instruction.

Subroutines contained within the program of memory 13 also cause a change in execution sequence, from a sequential instruction to a displaced instruction. The explicit prefetch instruction can be used when divergence instructions are encountered, such as "if " statements. Conditional instructions may also be preceded by a prefetch instruction to advantageously prefetch a non-sequential target instruction.

In selecting a location of a prefetch instruction, vis a vis, a branch instruction, such as to permit sufficient time for prefetching the required target instruction before execution of the branch instruction, the prefetch instruction should be located a number of instructions away from the branch instruction to permit ample prefetch time. In FIG. 2, this is illustrated as being only one address earlier.

It is also possible to include more than one prefetch instruction, identifying a second target from a branch instruction. This would then provide prefetched targets which would be available once the branch instruction is executed.

Overuse of the prefetch instructions for all branch instructions is to be avoided. It is clear that by inserting prefetch instructions that the branch target is prefetched ahead of execution, whether or not that branch target may be needed. This is additional bus cycle overhead to prefetch an instruction which may ultimately not be used.

It is clear that those skilled in the programming arts may selectively insert explicit cache prefetch instructions for those branch instructions which have a higher probability of having a cache miss for their branch targets. This would include branches which have a target with an absolute displacement from the branch instruction which is greater than the cache size. Employing the prefetch instruction for these conditions will reduce the probability of a cache miss without overburdening the bus 10 with instruction prefetches which are unnecessary. By observing the foregoing principle, that when a branch instruction requires transfer of program execution to an instruction having a displacement from the branch instruction greater than the cache size, the probability of having that branch target in cache is very low unless the instruction is explicitly fetched.

The foregoing techniques of providing explicit cache prefetching may be accomplished by using some hardware decoding logic in the instruction cache 12. By decoding the OPCODE of each prefetched instruction, it is possible to determine the presence of a branch instruction. When the OPCODE detection indicates that the instruction is part of an unconditional branch instruction, the sequential cache prefetch is stopped, and a prefetch of the branch target identified by the branch instruction is started. If the OPCODE is part of some condition branch instruction, instructions from different branch targets are prefetched according to any predetermined probability of executing those branches.

This predecoding of instructions prefetched in the instruction cache is to be distinguished from conventional techniques for prefetching branch instructions from a branch instruction currently being decoded and executed by the processor. In the prior art technique, any latency between detection of a branch OPCODE and actual use of the branch target would be very small. The present solution, however, proposes to inspect instructions which are in the instruction cache, before execution, prefetching branch targets based on the OPCODE decoding. Other techniques which take advantage of a pre-analysis of the program stored in the program memory 13 include selectively freezing the contents of the instruction cache 18. Using the foregoing sequential prefetching as the normal prefetch mode, it is possible to freeze elements of a block of instructions, in the instruction cache 12, which are likely to be used again. Rather than explicitly referencing back to these instructions, which were once used and then discarded, they are maintained in the instruction cache and may be read only and not replaced while the program is executing within the block.

Selective cache line freezing is useful in situations wherein the cache replacement policy is LRU (least recently used). In this circumstance, the last instruction to be used in the cache is replaced with the next prefetched instruction. Thus, when sequentially executing a block of instruction containing a loop having a size greater than the cache size, the loop instructions are continuously and sequentially fetched into the cache, referenced by the processor only once, and then immediately replaced before they are referenced again. This promotes an inefficient prefetching and demands on bus cycle time by constantly prefetching instructions which were reused only a few execution cycles earlier.

Referring to FIG. 3, there is shown a sequence of program instructions beginning with address 000 through address 100, which illustrates the process of freezing commonly used instructions. Contained within the block of program instructions of FIG. 3 is a branch instruction at address 010. By including a freeze bit with instruction 1, which may be a target instruction for the branch instruction at address 010, it is possible to hold this instruction in the cache and not permit it to be replaced until the program exits the loop to a higher numbered instruction. An unfreeze, explicit instruction may be located at address 011, which, after the loop has been completed within the block represented by instructions at addresses 001 through 010, permits these frozen instructions to be discarded on subsequent prefetch cycles.

FIG. 4 illustrates how instructions at addresses 001, 002 and 003 may be frozen in a portion of the cache, permitting prefetching only in lines 4, 5 and 6 of the cache. In this way, as the sequential prefetching of instructions from addresses 001 through 003 commence, they will be loaded into the cache and frozen. Instructions at addresses 004 through 009 will be prefetched, preferably two at a time, during subsequent available bus cycles, and executed by the processor. At the time of prefetching of the branch instruction at address 010, the instructions at addresses 001, 002 and 003 will have been frozen in the first three line portions of the cache. Thus, since 001 is a target of branch instruction at address 010, prefetching of the target instructions will not be necessary. Further, no prefetching will be necessary until instruction 003 has been executed, wherein the remaining sequential instructions at addresses 004 through 010 may be inserted in the unfrozen portion of the cache. Once the unfreeze instruction 011 has been referenced by the processor as a result of execution of the branch instruction, indicating that the loop is completed, the unfreeze instruction will permit all six (6) locations in the instruction cache to be used in sequential prefetching of further instructions.

To select which lines of a cache should be frozen, the following criteria should be considered.

The entry line of a basic block has the highest priority to be frozen in the cache. If the block of instructions includes a loop which has as its target the entry line of a basic block (a commonly encountered situation), no branch target prefetching will be necessary for loop execution of the entry line is frozen in the cache. By freezing the first one or two entry lines of a block of instructions, it is clear that when the processor references the target, i.e., the entry line, the instruction cache will already contain the requested instruction. This consequence will not only reduce bus traffic, but eliminate the need for explicit cache instruction to prefetch the branch target, or any extra hardware decoding for conducting OPCODE inspection.

The cache should be selectively frozen in such a way that the frozen cache lines and the implicit cache prefetching for the unfrozen portion of the cache are distributed uniformly across those instructions constituting a loop. The size of the frozen portion of the instruction cache should be as large as possible, being certain that the free unfrozen portion is sufficiently large to handle the remaining instructions of a loop.

FIG. 5 illustrates the architecture for implementing an instruction cache having a portion which can be frozen, i.e., only read and a portion which may be freely written and read. The processor 11 is shown having a program counter 18 which identifies the address of the next instruction to be executed. A prefetch unit 19, which is itself a processor, is shown which will supervise the prefetching of lines of instruction from memory 13, based upon the program counter 18 and the foregoing prefetching techniques.

Assuming that the entry instruction for a block of instructions has been identified by program counter 18 for the first time, i.e., a loop has not yet been entered, the prefetch unit 19 will determine whether or not the line has been previously prefetched, and is available in either the prefetch buffer 21, current buffer 20(b), or instruction cache 20(a). The prefetching unit 19 will generate a line request in response to the program counter 18 to be certain the next required instruction is available for the processor 11. The prefetching unit will initially check all the buffer units, i.e., P buffer 21, C buffer 20(b) and instruction cache 20(a) to determine whether a line containing the desired instruction is in one of these units. If there is a cache hit on any one of the buffer units, i.e., the requested instruction is present in these units, there is no prefetch necessary, and no request is made of the memory 13 for the instruction.

If a cache miss occurs, a prefetch request will be sent to the main memory on the next available time slot for bus 25. When the prefetched line arrives, which is normally the next sequential line or pairs of lines in a two line prefetch mode, they are inserted in the P buffer 21. The newly arrived lines replace the least recently used line of instructions, according to the LRU criteria.

When the processing unit issues the request for the next available instruction, as determined by program counter 18, all the buffers will be searched for the requisite instruction. If the requisite instruction is found on the prefetched buffer, the reference line is transferred to the current buffer. If the referenced line is found in the current buffer, all the contents of the buffers will remain the same. If the instruction requested is found in the instruction cache 20(a), the content of all the buffers remains unchanged.

The instruction cache 20(a) serves as a portion of the instruction cache which can be frozen. The free portion of the instruction cache is that denoted C buffer by 20(b).

The system of FIG. 5 will prefetch instructions from the memory 13 depending on the counter 18. When a hit is made by processor 11 on the P buffer 21, the referenced line is transferred to the current buffer. This will continue until the current buffer is full. Once the current buffer 20(b) is full, the LRU rule is invoked to transfer the least referenced instruction to the instruction cache 20(a) if the instruction cache is not frozen. If the instruction cache 20(a) is frozen, any line replaced from the C buffer 20(b) will not be put in instruction cache 20(a). Thus, once a freeze instruction has been executed by processor 11, the instruction cache 20(a) is frozen, thus retaining the previously fetched instructions therein. These will advantageously be the early lines of a loop of an instruction block or other such target instruction. The C buffer 20(b) is free to receive further prefetched instructions via the P buffer 21.

Thus, there are described several embodiments of an overall system for controlling prefetching of instructions from an instruction memory 13. These cache management techniques will reduce the required bus access time for prefetching instructions, while reducing the overall number of cache misses. Those skilled in the art will recognize yet other embodiments of the invention described more particularly by the claims which follow.

Claims (5)

What is claimed is:
1. In a processing system for executing program instructions stored in a memory and forming subgroups of blocks of instructions, at least one of said blocks including a group of instructions which are repeatedly executed, a system for improving execution efficiency of said program comprising:
an instruction cache for storing a minor number of instructions of a block of said instructions; and
a prefetch processor programmed to sequentially fetch a number of instructions of said program from the memory and store the number of instructions in said instruction cache, said processor controlling said cache to freeze a portion of said cache containing said instructions which are repeatedly executed, while prefetching sequentially remaining instructions of said program, said prefetch processor being operative to inspect instructions, fetched from the memory to control the freezing.
2. The system of claim 1 wherein said prefetch processor freezes the instructions representing the beginning portion of a block containing said repeatedly executed instructions.
3. The system of claim 1 wherein said repeatedly executed instructions are tagged, and said prefetch processor detects said tagged instructions and subsequently freezes said tagged instructions in said cache.
4. The system of claim 3 wherein said program includes an instruction which unfreezes said cache.
5. The system of claim 4 wherein all of said program instructions include a bit position which identifies an instruction as an instruction to be frozen or which is not to be frozen.
US08063845 1990-03-27 1993-05-19 Instruction cache system for implementing programs having non-sequential instructions and method of implementing same Expired - Lifetime US5701435A (en)

Priority Applications (2)

Application Number Priority Date Filing Date Title
US50061290 true 1990-03-27 1990-03-27
US08063845 US5701435A (en) 1990-03-27 1993-05-19 Instruction cache system for implementing programs having non-sequential instructions and method of implementing same

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
US08063845 US5701435A (en) 1990-03-27 1993-05-19 Instruction cache system for implementing programs having non-sequential instructions and method of implementing same

Related Parent Applications (1)

Application Number Title Priority Date Filing Date
US50061290 Continuation 1990-03-27 1990-03-27

Publications (1)

Publication Number Publication Date
US5701435A true US5701435A (en) 1997-12-23

Family

ID=23990184

Family Applications (1)

Application Number Title Priority Date Filing Date
US08063845 Expired - Lifetime US5701435A (en) 1990-03-27 1993-05-19 Instruction cache system for implementing programs having non-sequential instructions and method of implementing same

Country Status (4)

Country Link
US (1) US5701435A (en)
JP (1) JP3095802B2 (en)
DE (2) DE69129872T2 (en)
EP (1) EP0449369B1 (en)

Cited By (30)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5951679A (en) * 1996-10-31 1999-09-14 Texas Instruments Incorporated Microprocessor circuits, systems, and methods for issuing successive iterations of a short backward branch loop in a single cycle
US5958045A (en) * 1997-04-02 1999-09-28 Advanced Micro Devices, Inc. Start of access instruction configured to indicate an access mode for fetching memory operands in a microprocessor
US6016533A (en) * 1997-12-16 2000-01-18 Advanced Micro Devices, Inc. Way prediction logic for cache array
US6052776A (en) * 1996-10-18 2000-04-18 Hitachi, Ltd. Branch operation system where instructions are queued until preparations is ascertained to be completed and branch distance is considered as an execution condition
US6073230A (en) * 1997-06-11 2000-06-06 Advanced Micro Devices, Inc. Instruction fetch unit configured to provide sequential way prediction for sequential instruction fetches
US6393527B1 (en) * 1998-12-18 2002-05-21 Ati International Srl Prefetch buffer with continue detect
US20030046494A1 (en) * 2001-08-29 2003-03-06 Ziegler Michael L. Program control flow conditioned on presence of requested data in cache memory
US6748496B1 (en) * 2000-04-18 2004-06-08 Ati International Srl Method and apparatus for providing cacheable data to a peripheral device
US20050050278A1 (en) * 2003-09-03 2005-03-03 Advanced Micro Devices, Inc. Low power way-predicted cache
US20050050277A1 (en) * 2003-09-03 2005-03-03 Advanced Micro Devices, Inc. MicroTLB and micro TAG for reducing power in a processor
US20050066132A1 (en) * 2003-09-24 2005-03-24 Matsushita Electric Industrial Co., Ltd. Information processing control system
US7055000B1 (en) 2003-08-29 2006-05-30 Western Digital Technologies, Inc. Disk drive employing enhanced instruction cache management to facilitate non-sequential immediate operands
US7216204B2 (en) 2001-08-27 2007-05-08 Intel Corporation Mechanism for providing early coherency detection to enable high performance memory updates in a latency sensitive multithreaded environment
US7225281B2 (en) 2001-08-27 2007-05-29 Intel Corporation Multiprocessor infrastructure for providing flexible bandwidth allocation via multiple instantiations of separate data buses, control buses and support mechanisms
US7246197B2 (en) 2001-08-27 2007-07-17 Intel Corporation Software controlled content addressable memory in a general purpose execution datapath
US20080034187A1 (en) * 2006-08-02 2008-02-07 Brian Michael Stempel Method and Apparatus for Prefetching Non-Sequential Instruction Addresses
US7337275B2 (en) 2002-08-13 2008-02-26 Intel Corporation Free list and ring data structure management
EP1901169A1 (en) * 2006-09-14 2008-03-19 SiTel Semiconductor B.V. Arrangement with instruction cache
US7418571B2 (en) 2003-01-10 2008-08-26 Intel Corporation Memory interleaving
US7487505B2 (en) 2001-08-27 2009-02-03 Intel Corporation Multithreaded microprocessor with register allocation based on number of active threads
US7610451B2 (en) 2002-01-25 2009-10-27 Intel Corporation Data transfer mechanism using unidirectional pull bus and push bus
US7681018B2 (en) 2000-08-31 2010-03-16 Intel Corporation Method and apparatus for providing large register address space while maximizing cycletime performance for a multi-threaded register file set
US20100146495A1 (en) * 2008-12-10 2010-06-10 Sun Microsystems, Inc. Method and system for interprocedural prefetching
US20100332800A1 (en) * 2009-06-30 2010-12-30 Fujitsu Limited Instruction control device, instruction control method, and processor
US20110161632A1 (en) * 2007-12-30 2011-06-30 Tingting Sha Compiler assisted low power and high performance load handling
US7991983B2 (en) 1999-09-01 2011-08-02 Intel Corporation Register set used in multithreaded parallel processor architecture
US9471322B2 (en) 2014-02-12 2016-10-18 Apple Inc. Early loop buffer mode entry upon number of mispredictions of exit condition exceeding threshold
US9557999B2 (en) 2012-06-15 2017-01-31 Apple Inc. Loop buffer learning
US9753733B2 (en) 2012-06-15 2017-09-05 Apple Inc. Methods, apparatus, and processors for packing multiple iterations of loop in a loop buffer
US9772824B2 (en) * 2015-03-25 2017-09-26 International Business Machines Corporation Program structure-based blocking

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
GB2247546B (en) * 1990-08-31 1995-03-01 Sun Microsystems Inc Method and apparatus for predictive caching
US5226138A (en) * 1990-11-27 1993-07-06 Sun Microsystems, Inc. Method for selectively transferring data instructions to a cache memory
US5774164A (en) * 1994-10-27 1998-06-30 Dai Nippon Printing Co., Ltd. Thermal transfer image-receiving sheet
DE19747275A1 (en) * 1997-10-25 1999-04-29 Philips Patentverwaltung Mobile device with a control signal generator
WO2002061574A1 (en) 2001-01-30 2002-08-08 Koninklijke Philips Electronics N.V. Computer instruction with instruction fetch control bits

Citations (20)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4157587A (en) * 1977-12-22 1979-06-05 Honeywell Information Systems Inc. High speed buffer memory system with word prefetch
US4189770A (en) * 1978-03-16 1980-02-19 International Business Machines Corporation Cache bypass control for operand fetches
US4437149A (en) * 1980-11-17 1984-03-13 International Business Machines Corporation Cache memory architecture with decoding
US4500954A (en) * 1981-10-15 1985-02-19 International Business Machines Corporation Cache bypass system with post-block transfer directory examinations for updating cache and/or maintaining bypass
US4594659A (en) * 1982-10-13 1986-06-10 Honeywell Information Systems Inc. Method and apparatus for prefetching instructions for a central execution pipeline unit
US4626988A (en) * 1983-03-07 1986-12-02 International Business Machines Corporation Instruction fetch look-aside buffer with loop mode control
US4691277A (en) * 1984-10-24 1987-09-01 International Business Machines Corp. Small instruction cache using branch target table to effect instruction prefetch
US4761731A (en) * 1985-08-14 1988-08-02 Control Data Corporation Look-ahead instruction fetch control for a cache memory
US4802085A (en) * 1987-01-22 1989-01-31 National Semiconductor Corporation Apparatus and method for detecting and handling memory-mapped I/O by a pipelined microprocessor
US4807110A (en) * 1984-04-06 1989-02-21 International Business Machines Corporation Prefetching system for a cache having a second directory for sequentially accessed blocks
US4851993A (en) * 1987-04-20 1989-07-25 Amdahl Corporation Cache move-in bypass
US4860197A (en) * 1987-07-31 1989-08-22 Prime Computer, Inc. Branch cache system with instruction boundary determination independent of parcel boundary
US4876642A (en) * 1988-01-19 1989-10-24 Gibson Glenn A Rules and apparatus for a loop capturing code buffer that prefetches instructions
US4894772A (en) * 1987-07-31 1990-01-16 Prime Computer, Inc. Method and apparatus for qualifying branch cache entries
US4937738A (en) * 1984-08-31 1990-06-26 Hitachi, Ltd. Data processing system which selectively bypasses a cache memory in fetching information based upon bit information of an instruction
US4942518A (en) * 1984-06-20 1990-07-17 Convex Computer Corporation Cache store bypass for computer
US4943908A (en) * 1987-12-02 1990-07-24 International Business Machines Corporation Multiple branch analyzer for prefetching cache lines
US5025366A (en) * 1988-01-20 1991-06-18 Advanced Micro Devices, Inc. Organization of an integrated cache unit for flexible usage in cache system design
US5099419A (en) * 1988-11-25 1992-03-24 Nec Corporation Pipeline microcomputer having branch instruction detector and bus controller for producing and carrying branch destination address prior to instruction execution
US5113370A (en) * 1987-12-25 1992-05-12 Hitachi, Ltd. Instruction buffer control system using buffer partitions and selective instruction replacement for processing large instruction loops

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US3466613A (en) * 1967-01-13 1969-09-09 Ibm Instruction buffering system

Patent Citations (20)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4157587A (en) * 1977-12-22 1979-06-05 Honeywell Information Systems Inc. High speed buffer memory system with word prefetch
US4189770A (en) * 1978-03-16 1980-02-19 International Business Machines Corporation Cache bypass control for operand fetches
US4437149A (en) * 1980-11-17 1984-03-13 International Business Machines Corporation Cache memory architecture with decoding
US4500954A (en) * 1981-10-15 1985-02-19 International Business Machines Corporation Cache bypass system with post-block transfer directory examinations for updating cache and/or maintaining bypass
US4594659A (en) * 1982-10-13 1986-06-10 Honeywell Information Systems Inc. Method and apparatus for prefetching instructions for a central execution pipeline unit
US4626988A (en) * 1983-03-07 1986-12-02 International Business Machines Corporation Instruction fetch look-aside buffer with loop mode control
US4807110A (en) * 1984-04-06 1989-02-21 International Business Machines Corporation Prefetching system for a cache having a second directory for sequentially accessed blocks
US4942518A (en) * 1984-06-20 1990-07-17 Convex Computer Corporation Cache store bypass for computer
US4937738A (en) * 1984-08-31 1990-06-26 Hitachi, Ltd. Data processing system which selectively bypasses a cache memory in fetching information based upon bit information of an instruction
US4691277A (en) * 1984-10-24 1987-09-01 International Business Machines Corp. Small instruction cache using branch target table to effect instruction prefetch
US4761731A (en) * 1985-08-14 1988-08-02 Control Data Corporation Look-ahead instruction fetch control for a cache memory
US4802085A (en) * 1987-01-22 1989-01-31 National Semiconductor Corporation Apparatus and method for detecting and handling memory-mapped I/O by a pipelined microprocessor
US4851993A (en) * 1987-04-20 1989-07-25 Amdahl Corporation Cache move-in bypass
US4894772A (en) * 1987-07-31 1990-01-16 Prime Computer, Inc. Method and apparatus for qualifying branch cache entries
US4860197A (en) * 1987-07-31 1989-08-22 Prime Computer, Inc. Branch cache system with instruction boundary determination independent of parcel boundary
US4943908A (en) * 1987-12-02 1990-07-24 International Business Machines Corporation Multiple branch analyzer for prefetching cache lines
US5113370A (en) * 1987-12-25 1992-05-12 Hitachi, Ltd. Instruction buffer control system using buffer partitions and selective instruction replacement for processing large instruction loops
US4876642A (en) * 1988-01-19 1989-10-24 Gibson Glenn A Rules and apparatus for a loop capturing code buffer that prefetches instructions
US5025366A (en) * 1988-01-20 1991-06-18 Advanced Micro Devices, Inc. Organization of an integrated cache unit for flexible usage in cache system design
US5099419A (en) * 1988-11-25 1992-03-24 Nec Corporation Pipeline microcomputer having branch instruction detector and bus controller for producing and carrying branch destination address prior to instruction execution

Non-Patent Citations (12)

* Cited by examiner, † Cited by third party
Title
Alpert; "Performance Trade-offs for Microprocessor Cache Memories" IEEE Micro Aug. 1988.
Alpert; Performance Trade offs for Microprocessor Cache Memories IEEE Micro Aug. 1988. *
Chi et al. "Compiler-Driver Cache Policy"; Jun. 1987, Purdue Univ.
Chi et al. "Improving Cache Performance by Selective Cache Bypass"; IEEE Jan. 1989.
Chi et al. Compiler Driver Cache Policy ; Jun. 1987, Purdue Univ. *
Chi et al. Improving Cache Performance by Selective Cache Bypass ; IEEE Jan. 1989. *
Cytron et al.; "Automatic Management of Programable Caches" IEEE Aug. 1988.
Cytron et al.; Automatic Management of Programable Caches IEEE Aug. 1988. *
Furht et al; "A Compiler Directed Cache Strategy for GaAs Microprocessor Architecture" Mar. 1988; IEEE.
Furht et al; A Compiler Directed Cache Strategy for GaAs Microprocessor Architecture Mar. 1988; IEEE. *
McNiven et al. "Analysis of Memory Referencing Behavior For Design of Local Memories" IEEE 1988.
McNiven et al. Analysis of Memory Referencing Behavior For Design of Local Memories IEEE 1988. *

Cited By (41)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6052776A (en) * 1996-10-18 2000-04-18 Hitachi, Ltd. Branch operation system where instructions are queued until preparations is ascertained to be completed and branch distance is considered as an execution condition
US5951679A (en) * 1996-10-31 1999-09-14 Texas Instruments Incorporated Microprocessor circuits, systems, and methods for issuing successive iterations of a short backward branch loop in a single cycle
US5958045A (en) * 1997-04-02 1999-09-28 Advanced Micro Devices, Inc. Start of access instruction configured to indicate an access mode for fetching memory operands in a microprocessor
US6101595A (en) * 1997-06-11 2000-08-08 Advanced Micro Devices, Inc. Fetching instructions from an instruction cache using sequential way prediction
US6073230A (en) * 1997-06-11 2000-06-06 Advanced Micro Devices, Inc. Instruction fetch unit configured to provide sequential way prediction for sequential instruction fetches
US6016533A (en) * 1997-12-16 2000-01-18 Advanced Micro Devices, Inc. Way prediction logic for cache array
US6115792A (en) * 1997-12-16 2000-09-05 Advanced Micro Devices, Inc. Way prediction logic for cache array
US6393527B1 (en) * 1998-12-18 2002-05-21 Ati International Srl Prefetch buffer with continue detect
US7991983B2 (en) 1999-09-01 2011-08-02 Intel Corporation Register set used in multithreaded parallel processor architecture
US6748496B1 (en) * 2000-04-18 2004-06-08 Ati International Srl Method and apparatus for providing cacheable data to a peripheral device
US7743235B2 (en) 2000-08-31 2010-06-22 Intel Corporation Processor having a dedicated hash unit integrated within
US7681018B2 (en) 2000-08-31 2010-03-16 Intel Corporation Method and apparatus for providing large register address space while maximizing cycletime performance for a multi-threaded register file set
US7216204B2 (en) 2001-08-27 2007-05-08 Intel Corporation Mechanism for providing early coherency detection to enable high performance memory updates in a latency sensitive multithreaded environment
US7487505B2 (en) 2001-08-27 2009-02-03 Intel Corporation Multithreaded microprocessor with register allocation based on number of active threads
US7225281B2 (en) 2001-08-27 2007-05-29 Intel Corporation Multiprocessor infrastructure for providing flexible bandwidth allocation via multiple instantiations of separate data buses, control buses and support mechanisms
US7246197B2 (en) 2001-08-27 2007-07-17 Intel Corporation Software controlled content addressable memory in a general purpose execution datapath
US6925535B2 (en) * 2001-08-29 2005-08-02 Hewlett-Packard Development Company, L.P. Program control flow conditioned on presence of requested data in cache memory
US20030046494A1 (en) * 2001-08-29 2003-03-06 Ziegler Michael L. Program control flow conditioned on presence of requested data in cache memory
US7610451B2 (en) 2002-01-25 2009-10-27 Intel Corporation Data transfer mechanism using unidirectional pull bus and push bus
US7337275B2 (en) 2002-08-13 2008-02-26 Intel Corporation Free list and ring data structure management
US7418571B2 (en) 2003-01-10 2008-08-26 Intel Corporation Memory interleaving
US7055000B1 (en) 2003-08-29 2006-05-30 Western Digital Technologies, Inc. Disk drive employing enhanced instruction cache management to facilitate non-sequential immediate operands
US20050050278A1 (en) * 2003-09-03 2005-03-03 Advanced Micro Devices, Inc. Low power way-predicted cache
US20050050277A1 (en) * 2003-09-03 2005-03-03 Advanced Micro Devices, Inc. MicroTLB and micro TAG for reducing power in a processor
US7117290B2 (en) 2003-09-03 2006-10-03 Advanced Micro Devices, Inc. MicroTLB and micro tag for reducing power in a processor
US20050066132A1 (en) * 2003-09-24 2005-03-24 Matsushita Electric Industrial Co., Ltd. Information processing control system
US8135909B2 (en) * 2003-09-24 2012-03-13 Panasonic Corporation System for starting a preload of a second program while a first program is executing
US7917731B2 (en) * 2006-08-02 2011-03-29 Qualcomm Incorporated Method and apparatus for prefetching non-sequential instruction addresses
US20080034187A1 (en) * 2006-08-02 2008-02-07 Brian Michael Stempel Method and Apparatus for Prefetching Non-Sequential Instruction Addresses
EP1901169A1 (en) * 2006-09-14 2008-03-19 SiTel Semiconductor B.V. Arrangement with instruction cache
WO2008033020A1 (en) 2006-09-14 2008-03-20 Sitel Semiconductor B.V. Arrangement with instruction cache
US20110161632A1 (en) * 2007-12-30 2011-06-30 Tingting Sha Compiler assisted low power and high performance load handling
US9311085B2 (en) * 2007-12-30 2016-04-12 Intel Corporation Compiler assisted low power and high performance load handling based on load types
US8312442B2 (en) * 2008-12-10 2012-11-13 Oracle America, Inc. Method and system for interprocedural prefetching
US20100146495A1 (en) * 2008-12-10 2010-06-10 Sun Microsystems, Inc. Method and system for interprocedural prefetching
US20100332800A1 (en) * 2009-06-30 2010-12-30 Fujitsu Limited Instruction control device, instruction control method, and processor
US9557999B2 (en) 2012-06-15 2017-01-31 Apple Inc. Loop buffer learning
US9753733B2 (en) 2012-06-15 2017-09-05 Apple Inc. Methods, apparatus, and processors for packing multiple iterations of loop in a loop buffer
US9471322B2 (en) 2014-02-12 2016-10-18 Apple Inc. Early loop buffer mode entry upon number of mispredictions of exit condition exceeding threshold
US9772824B2 (en) * 2015-03-25 2017-09-26 International Business Machines Corporation Program structure-based blocking
US9772825B2 (en) * 2015-03-25 2017-09-26 International Business Machines Corporation Program structure-based blocking

Also Published As

Publication number Publication date Type
DE69129872D1 (en) 1998-09-03 grant
JP3095802B2 (en) 2000-10-10 grant
EP0449369A3 (en) 1992-03-18 application
EP0449369A2 (en) 1991-10-02 application
DE69129872T2 (en) 1999-03-04 grant
EP0449369B1 (en) 1998-07-29 grant
JPH04225428A (en) 1992-08-14 application

Similar Documents

Publication Publication Date Title
Eickemeyer et al. A load-instruction unit for pipelined processors
Kessler et al. The Alpha 21264 microprocessor architecture
US5805877A (en) Data processor with branch target address cache and method of operation
US5732242A (en) Consistently specifying way destinations through prefetching hints
US6697932B1 (en) System and method for early resolution of low confidence branches and safe data cache accesses
US5850543A (en) Microprocessor with speculative instruction pipelining storing a speculative register value within branch target buffer for use in speculatively executing instructions after a return
US4725947A (en) Data processor with a branch target instruction storage
US6289442B1 (en) Circuit and method for tagging and invalidating speculatively executed instructions
US6295600B1 (en) Thread switch on blocked load or store using instruction thread field
US5506976A (en) Branch cache
US5832297A (en) Superscalar microprocessor load/store unit employing a unified buffer and separate pointers for load and store operations
US6178498B1 (en) Storing predicted branch target address in different storage according to importance hint in branch prediction instruction
US5687360A (en) Branch predictor using multiple prediction heuristics and a heuristic identifier in the branch instruction
US5566324A (en) Computer apparatus including a main memory prefetch cache and method of operation thereof
US6157988A (en) Method and apparatus for high performance branching in pipelined microsystems
US6539458B2 (en) Hierarchical memory for efficient data exchange control
US4974154A (en) Computer with instruction prefetch queue retreat unit
US6134633A (en) Prefetch management in cache memory
US5717895A (en) Associative scalar data cache with write-through capabilities for a vector processor
US6401193B1 (en) Dynamic data prefetching based on program counter and addressing mode
US4775927A (en) Processor including fetch operation for branch instruction with control tag
US6427192B1 (en) Method and apparatus for caching victimized branch predictions
US5265213A (en) Pipeline system for executing predicted branch target instruction in a cycle concurrently with the execution of branch instruction
US5944815A (en) Microprocessor configured to execute a prefetch instruction including an access count field defining an expected number of access
US6263428B1 (en) Branch predictor

Legal Events

Date Code Title Description
AS Assignment

Owner name: U.S. PHILIPS CORPORATION, NEW YORK

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:PHILIPS ELECTRONICS NORTH AMERICA CORPORATION;REEL/FRAME:008698/0171

Effective date: 19970821

FPAY Fee payment

Year of fee payment: 4

FPAY Fee payment

Year of fee payment: 8

AS Assignment

Owner name: NXP B.V., NETHERLANDS

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:PHILIPS ELECTRONICS NORTH AMERICA CORP.;REEL/FRAME:018654/0514

Effective date: 20061206

Owner name: PHILIPS ELECTRONICS NORTH AMERICAN CORPORATION, NE

Free format text: CHANGE OF NAME;ASSIGNOR:NORTH AMERICAN PHILIPS CORPORATION;REEL/FRAME:018654/0565

Effective date: 19930524

AS Assignment

Owner name: MORGAN STANLEY SENIOR FUNDING, INC., ENGLAND

Free format text: SECURITY AGREEMENT;ASSIGNOR:NXP B.V.;REEL/FRAME:018806/0201

Effective date: 20061201

FPAY Fee payment

Year of fee payment: 12

AS Assignment

Owner name: NXP B.V., NETHERLANDS

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:U.S. PHILIPS CORPORATION;REEL/FRAME:024927/0151

Effective date: 20100901

AS Assignment

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:CHI, CHI-HUNG;REEL/FRAME:025017/0535

Effective date: 19900327

Owner name: NORTH AMERICAN PHILIPS CORPORATION, NEW YORK

AS Assignment

Effective date: 20101122

Owner name: NXP B.V., NETHERLANDS

Free format text: NUNC PRO TUNC ASSIGNMENT;ASSIGNOR:U.S. PHILIPS CORPORATION;REEL/FRAME:025392/0945

AS Assignment

Free format text: RELEASE BY SECURED PARTY;ASSIGNOR:MORGAN STANLEY SENIOR FUNDING, INC.;REEL/FRAME:026153/0847

Effective date: 20101229

Owner name: NXP B.V., NETHERLANDS

AS Assignment

Effective date: 20110407

Owner name: LIYUE CAPITAL, LLC, DELAWARE

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:SINO MATRIX TECHNOLOGY INC.;REEL/FRAME:026160/0125

Owner name: SINO MATRIX TECHNOLOGY INC., TAIWAN

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:NXP B.V.;REEL/FRAME:026160/0111

Effective date: 20110225

AS Assignment

Effective date: 20060929

Owner name: MORGAN STANLEY SENIOR FUNDING, INC., UNITED KINGDO

Free format text: CORRECTION TO THE RECORDATION COVER SHEET OF THE SECURITY AGREEMENT RECORDED AT 018806/0201 ON 1/22/2007;ASSIGNOR:NXP B.V.;REEL/FRAME:026319/0879

AS Assignment

Free format text: MERGER;ASSIGNOR:LIYUE CAPITAL, LLC;REEL/FRAME:031369/0368

Owner name: INTELLECTUAL VENTURES II LLC, DELAWARE

Effective date: 20131007